On 2009-07-13, Andres Freund <andres@anarazel.de> wrote:
> On Sunday 12 July 2009 13:19:50 Phoenix Kiula wrote:
>> Hi. I *always* get an error moving my current fully utf-8 database
>> data into a new DB.
>>
>> My server has the version 8.3 with a five year old DB. Everything, all
>> collation, LC_LOCALE etc are all utf8.
>>
>> When I install a new Postgresql 8.4 on my home Mac OSX machine (after
>> losing some hair) I set everything about a new database to be utf8. At
>> least anything I could see in PgAdmin.
>>
>> But when I pull in the data dump from the server I always see that
>> error from the utf8 mismatch and such.
>>
>> So, my question. What is a good way to make sure that error does NOT
>> occur? I simply wish to replicate the server database on another PG
>> installation. What should one do?
> What is the _exact_ error you get? During which statement?
>
> Andres
>
if you do an ascii dump and the dump starts out "SET CLIENT ENCODING 'UTF8'"
or similar but you still get errors.
run it through run it through "iconv -f UFT8 -t UTF8//IGNORE"
that'll drop any illegal symbols. In theory that's a reduction in data
integrity.
iconv doesn't seem to have an option to replace them with U+FFFD :(
I had hoped that //TRASNSLIT would do that, but no.
